Re: [PATCH v2 1/6] powerpc/smp: Cache CPU has Asymmetric SMP
From: Michael Ellerman <mpe@ellerman.id.au>
Date: 2023-10-19 04:33:36
Also in:
lkml
Srikar Dronamraju [off-list ref] writes:
Currently cpu feature flag is checked whenever powerpc_smt_flags gets called. This is an unnecessary overhead. CPU_FTR_ASYM_SMT is set based on the processor and all processors will either have this set or will have it unset.
The cpu_has_feature() test is implemented with a static key. So AFAICS this is just replacing one static key with another? I see that you use the new static key in subsequent patches. But couldn't those just use the existing cpu feature test? Anyway I'd be interested to see how the generated code differs before/after this. cheers
quoted hunk ↗ jump to hunk
Hence only check for the feature flag once and cache it to be used subsequently. This commit will help avoid a branch in powerpc_smt_flags Signed-off-by: Srikar Dronamraju <redacted> --- Changelog: v1->v2: Using static keys instead of a variable. Using pr_info_once instead of printk arch/powerpc/kernel/smp.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-)diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c index 5826f5108a12..37c41297c9ce 100644 --- a/arch/powerpc/kernel/smp.c +++ b/arch/powerpc/kernel/smp.c@@ -988,18 +988,16 @@ static int __init init_thread_group_cache_map(int cpu, int cache_property) } static bool shared_caches; +DEFINE_STATIC_KEY_FALSE(powerpc_asym_packing); #ifdef CONFIG_SCHED_SMT /* cpumask of CPUs with asymmetric SMT dependency */ static int powerpc_smt_flags(void) { - int flags = SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES; + if (static_branch_unlikely(&powerpc_asym_packing)) + return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES | SD_ASYM_PACKING; - if (cpu_has_feature(CPU_FTR_ASYM_SMT)) { - printk_once(KERN_INFO "Enabling Asymmetric SMT scheduling\n"); - flags |= SD_ASYM_PACKING; - } - return flags; + return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES; } #endif@@ -1686,6 +1684,11 @@ static void __init fixup_topology(void) { int i; + if (cpu_has_feature(CPU_FTR_ASYM_SMT)) { + pr_info_once("Enabling Asymmetric SMT scheduling\n"); + static_branch_enable(&powerpc_asym_packing); + } + #ifdef CONFIG_SCHED_SMT if (has_big_cores) { pr_info("Big cores detected but using small core scheduling\n");-- 2.31.1